“By the time you do 70 acquisitions, you’ve made every mistake in the book.”
Mike Volpi (@mavolpi) did his first deal a month after joining Cisco. Years later, he was the first person our CEO Aaron Katz (@ceo_clickhouse) called about ClickHouse.
On episode 2 of Real Time, Mike joins Aaron Katz (@ceo_clickhouse) to talk about how he decides which risks are worth taking, from buying companies at Cisco to backing ClickHouse and investing in AI at @hanabicapital
A few highlights ⬇️
• Aaron reflects on the early days of building ClickHouse when they had 15 employees, no paying customers, and the market had just contracted. Mike explains how they navigated with a strong team including Benchmark’s (@ Peter Fenton)
• Why the AI boom reminds Mike of the early days of the internet. “You can’t sit by the sidelines when the best game in the world is being played.”
• The call Mike took in the middle of a board meeting about Ferrari’s electric car.
Timestamps:
00:00 Welcome to Real Time
02:43 Joining Cisco before the internet took off
05:56 Turning M&A into a business process at Cisco
11:17 The dot-com crash and running Cisco's biggest product lines
13:10 Leaving Cisco and lessons as a startup CEO
17:21 Building Index Ventures in the US
21:34 Venture capital and the open source thread
29:34 Backing ClickHouse with Peter Fenton
33:13 A smaller slice of a bigger pie
34:42 Why ClickHouse has to become more than a great database
38:14 Starting Hanabi Capital
43:56 Ferrari, scarcity and taking risks
51:56 Closing thoughts
Watch on YouTube or listen on Spotify and Apple Podcasts. Links to the full episode in the first reply ⬇️
Live sports won't wait for a batch job.
@SonyLIV uses ClickHouse Cloud to analyze billions of streaming events a day, helping teams spot buffering, playback failures and CDN issues while the action is live.
Try ClickHouse Cloud: https://t.co/Vl3vLCB6iN
Your coding agents need their own database.
An agent writing a schema migration has to test it against real data. Put several agents on one shared staging database and they start breaking each other’s tests.
Xata solves this with Postgres branches. Each branch is an isolated copy of your database, ready in seconds. Copy-on-write storage means a branch shares unchanged data with its parent, so every agent starts with the real schema and data without waiting for a full copy. Idle branches scale compute to zero, so you only pay for branches that are working.
Following our acquisition of Xata, we’re bringing branching into ClickHouse Managed Postgres.
On Monday, @saisrirampur and Xata CTO @tudor_g will show where the service goes next, with live demos and time for your questions.
Monday, Nov 23 · 10am PT / 1pm ET
Save your seat → https://t.co/vLtKb5OE6S
@mavolpi@ceo_clickhouse Watch on YouTube: https://t.co/EemgVHNU3m
Listen on Spotify: https://t.co/oJyzuJ2FBc
Listen on Apple Podcasts: https://t.co/WSfumd6cON
“By the time you do 70 acquisitions, you’ve made every mistake in the book.”
Mike Volpi (@mavolpi) did his first deal a month after joining Cisco. Years later, he was the first person our CEO Aaron Katz (@ceo_clickhouse) called about ClickHouse.
On episode 2 of Real Time, Mike joins Aaron Katz (@ceo_clickhouse) to talk about how he decides which risks are worth taking, from buying companies at Cisco to backing ClickHouse and investing in AI at @hanabicapital
A few highlights ⬇️
• Aaron reflects on the early days of building ClickHouse when they had 15 employees, no paying customers, and the market had just contracted. Mike explains how they navigated with a strong team including Benchmark’s (@ Peter Fenton)
• Why the AI boom reminds Mike of the early days of the internet. “You can’t sit by the sidelines when the best game in the world is being played.”
• The call Mike took in the middle of a board meeting about Ferrari’s electric car.
Timestamps:
00:00 Welcome to Real Time
02:43 Joining Cisco before the internet took off
05:56 Turning M&A into a business process at Cisco
11:17 The dot-com crash and running Cisco's biggest product lines
13:10 Leaving Cisco and lessons as a startup CEO
17:21 Building Index Ventures in the US
21:34 Venture capital and the open source thread
29:34 Backing ClickHouse with Peter Fenton
33:13 A smaller slice of a bigger pie
34:42 Why ClickHouse has to become more than a great database
38:14 Starting Hanabi Capital
43:56 Ferrari, scarcity and taking risks
51:56 Closing thoughts
Watch on YouTube or listen on Spotify and Apple Podcasts. Links to the full episode in the first reply ⬇️
“You can’t sit by the sidelines when the best game in the world is being played.”
GP at Hanabi Capital, Mike Volpi (@mavolpi), joins our CEO Aaron Katz (@ceo_clickhouse) for episode 2 of Real Time.
Here's a first look. The full episode is out today 12:30pm ET.
Your anomaly detector should know Christmas is coming. 🎄
In our ClickHouse + TimesFM demo, a Christmas flag brought 9 of 10 stores below the unchanged anomaly threshold. See how to supply that context through SQL.
https://t.co/cxKgvS2VYD
November brought our biggest launch since ClickHouse Cloud. 🗞️
Catch up on Managed Postgres GA, ClickHouse Observability, ClickHouse Security and the Langfuse AI Gateway, plus JSONBench results across 27 engines.
https://t.co/ze9ZyGaZaK
PostgreSQL 19 is out today, and you can run it on ClickHouse Managed Postgres from day one 🐘
No preview, no waitlist. Spin up a PG 19 instance right now: https://t.co/cToBP3BqvY
Day-one support for a new major version doesn’t happen by accident, and we’ll share the engineering story behind it soon.
🚨 Hackathon Announcement
On December 1, @langfuse and Kuro are hosting the Blueprint Hack in Berlin, an evening around building AI for construction.
Construction runs on messy visual data, and that makes it one of the hardest (and most fun) problems to throw AI at.
You’ll learn the language of construction, turn real drawings into real numbers, and compete for €2,000 in AI credits. Beer, pizza and tokens are on us.
There are 30 spots.
The ClickHouse Arrow Flight connector for Power BI is in public preview
Connect directly from the Power BI service, with no gateway to deploy. Data stays columnar end to end
Learn more https://t.co/Sad8RjnIEh
When an NVMe drive failed, ClickHouse writes started failing while Kubernetes still reported the node as Ready. 🔍
That incident led us to build a consistent node reliability system across three clouds.
https://t.co/jQNCf8UujH
pg_clickhouse 0.11.1 is out 🚀
The release adds pgvector support, with ClickHouse Array(Float32) columns mapped to the vector type and distance operators pushed down to ClickHouse, vector_similarity indexes included. UPDATE, DELETE and TRUNCATE push down to ClickHouse too, alongside regr_* aggregates, JSON subscripts and TimeZone handling.
Available on all ClickHouse Managed Postgres servers. Blog: https://t.co/qKKe1WDJ6a
For years, Front relied on Elasticsearch for customer-facing analytics.
But as data volumes expanded and the team’s ambitions grew, the engineering cost became unsustainable.
See why they migrated @ClickHouseDB for “pixel-perfect” analytics at scale 👇
https://t.co/rWtlK2PrVM
Trunk cut flaky test detection from up to an hour to five minutes, and observability costs 15-20x, by moving from TimescaleDB and Datadog to @ClickHouseDB.
See how one technology now powers their product analytics and agent-assisted investigations 👇
https://t.co/l6MCQkmwGX
Postgres for your application, ClickHouse for real-time analytics.
For something like an orders dashboard, the useful question is how those two fit together. How does the data reach the analytics side, and what does your team need to manage?
ClickHouse Managed Postgres has native integration with ClickHouse. On Monday, we'll show new product features live and talk about what we're building next.
If you're using Postgres today and thinking about adding analytics, bring the questions you have about your own application.
Monday, Nov 23 · 10am PT / 1pm ET
https://t.co/vLtKb5OE6S
1B+ LLM requests a day. One data foundation.
@OpenRouter uses ClickHouse Cloud to turn request telemetry into usage charts, endpoint health, model rankings and routing decisions.
Build on ClickHouse Cloud: https://t.co/Palop9V3vR
A ClickHouse Managed Postgres user reported high read replica lag. The cause was not enough load.
With our default settings, Postgres switches WAL segments every 16 MB or 60 seconds, whichever comes first. Under moderate write traffic, the 16 MB fills fast and the replica trails by a couple of seconds. On a quiet system, nothing ships for up to a minute.
We put it to the test with pgbench, a 200 ms heartbeat, and a lot of tuning:
https://t.co/OOfuq4Aesf